Biogen’s hopes for a quick regulatory approval of its anti-Aβ antibody aducanumab may be dimming. An advisory panel convened by the U.S. Food and Drug Administration was unimpressed by the evidence for efficacy. In a seven-hour virtual meeting today, the panel of neurologists and biostatisticians picked apart Biogen’s case, pointing to discrepancies in the dataset. Not a single one of the 11-person panel considered the evidence sufficient for approval.

The application rested on the results from two Phase 3 trials, EMERGE and ENGAGE, which were halted after failing an interim futility analysis (Mar 2019 news). After analyzing additional data, Biogen shocked the field by announcing one of the two trials, EMERGE, was positive after all (Oct 2019 news). In post hoc analyses, Biogen researchers then argued that the failure of ENGAGE could be explained by a lower total exposure, due to a mid-trial dose increase. Even then, Alzheimer’s researchers had mixed reactions to this analysis (Dec 2019 news). 

Biogen submitted its licensing application to the FDA four months ago; application to the European Medicines Agency is in the works (Jul 2020 news). 

The biologics license application made the case that the EMERGE results were strong enough to form the basis for approval, and that the previous positive Phase 1b PRIME trial could serve as supporting evidence (Mar 2015 conference news; Sep 2016 news). Biogen and the FDA's neurologists both argued, essentially, that the negative ENGAGE trial did not detract from the positive EMERGE trial. This was the case, they said, not only because of the lower exposure to the 10 mg/kg high dose in ENGAGE, but also because a larger number of fast progressors in the high-dose ENGAGE group essentially drowned out an emerging treatment signal.

The panelists did not buy it. With one positive and one negative trial, the positive result is as likely to be the anomaly as the negative result, they noted. They even bemoaned an inherent bias in the briefing document toward aducanumab’s efficacy.

The panelists agreed that the evidence for aducanumab removing amyloid plaques was strong, and commended Biogen for pushing the program forward. At the same time, they questioned whether the data gathered thus far showed that this amyloid removal was meaningful for efficacy, noting a weak correlation between plaque clearance and the slowing of cognitive decline, and a paucity of biomarker and pathophysiology evidence downstream of plaque removal.

The panel praised the FDA’s biostatistical review of the aducanumab filing, which found numerous holes in the data. In contrast, the FDA’s internal neurology review favored approval. The panel expressed consternation at the profound disconnect between the two.

The FDA collaborated with Biogen in preparing this licensing application. In general, the panel welcomed such collaboration in the interest of finding better treatments for Alzheimer’s disease, but also expressed concern at how close the working relationship appears to have been in this case.

The agency is under pressure to approve aducanumab from some advocacy groups and patients, several of whom spoke movingly at the meeting. The Alzheimer’s Association issued a statement urging approval. In contrast, the global advocacy group Alzheimer Disease International put out a neutral statement about awaiting the panel’s decision “with bated breath.” A letter from the American Academy of Neurology urged careful consideration of the logistic and financial consequences of approving aducanumab now, but took no position on the data package itself.

The FDA usually follows the guidance from its advisory panels, but is not obligated to do so. A decision will come by March 2021.—Madolyn Bowman Rogers

Comments

  1. This was the most unusual Advisory Committee meeting we've ever listened to. The opinions that were expressed were extremely positive or negative across the board, with very few participants having both positive and negative feedback. The complete disconnect between the clinical and statistical interpretations within the FDA was surprising and unexpected.

    The data and statistical interpretations are usually the foundation for the clinical interpretation, and clinical perspectives on the data are usually statistically supported within the FDA. This lack of alignment cannot become a trend or it will cause inconsistencies in the regulatory process that will rob the community of effective treatment, either by blocking good drugs or allowing bad ones through.

    There is no question that some of the data is consistent with an efficacy signal, but without an analysis combining primary and secondary efficacy outcomes, it is hard to know whether this signal is robust across subgroups or is driven by a small group of patients. With this analysis, and assuming results are consistent across subgroups, the EMERGE study is strong and the ENGAGE study may have provided supportive evidence. Biogen’s reliance and the Advisory Committee’s focus on the PRIME study was a distraction—ascending dose studies can't be interpreted the same as parallel group RCTs.

    The FDA decision will require answering some of these questions to at least get internal alignment and to thoroughly investigate the issues discussed by the advisors. The conclusion is likely to be as divisive as the presidential election, no matter which way it goes.

  2. Time to think outside the box

    Common sense has prevailed. The failing approach to AD therapy we have implemented so far reflects the faulty attitude we have toward neurodegeneration (Hurtley and Alderton, 2020). A growing body of evidence supports the notion that AD, Parkinson's disease (PD), and amyotrophic lateral sclerosis (ALS), to name the most frequent neurodegenerative conditions, share key pathogenic processes.

    In any of these conditions, no mono-causal treatment has so far reached its primary clinical endpoints, most likely because the dominating pathogenic constructs are too simplistic, not accounting for the complexity and variety of pathological processes that promote the neuronal loss and the expression of the clinical phenotypes (Espay et al., 2019). A wealth of evidence has instead indicated that the neurodegenerative processes set in motion in AD, PD, and ALS result from synergistic activities of many polygenic, epigenetic, environmental, vascular, and metabolic factors. Many AD patients exhibit mixed neuropathology and the presence of a different array of neurotoxic proteins, including Aβ, Tau, prion proteins (PrP), α-synuclein, and TAR DNA binding protein-43 (TDP-43) (Boyle et al., 2018; Boyle et al., 2019; Yu et al., 2018Karanth et al., 2020). This neuropathological mix is the rule, not the exception in AD-related neurodegeneration. Thus, it seems likely that a neurodegenerative "final common pathway" sets in motion the neuronal and non-neuronal death associated with the diseases. However, misfolded proteins are not necessarily, or at least not only, the molecular switches that act up- or downstream of the "common pathway" to swirl the process toward the expression of specific clinical phenotypes.

    The epistemological dead-end in which we are currently entrapped is the consequence of a shortcoming of modern medicine: the lack of systems-medicine- and network-based approaches to chronic diseases, as we instead persist on an organo-centric view that suffers from a reductionist bias (Loscalzo and Barabasi, 2011). 

    The consequences of such a view are extremely detrimental. The failure to consider the disease state due to the converging failure of multiple health-controlling systems and networks leads to the illusion that a magic "silver bullet" can provide the cure. On the contrary, great efforts are now required to explore the molecular granularity of these condition subtypes (Espay et al., 2019), because capturing and framing the complexity of these intrinsic and extrinsic factors represent an unmet clinical challenge. This challenge can be addressed by taking advantage of unbiased models of disease offered by artificial intelligence (AI) and machine learning (ML). ML techniques and big data analysis are becoming mandatory to identify novel and unexpected disease features.

    Thus, the search for a cure will likely depend on the willingness to explore unchartered territories and/or apply an "outside-the-box" line of thinking—perhaps, something similar to the successful "man on the moon" strategy because in less than two decades and, quite remarkably, the implementation of hypothesis-free and data-driven approaches, allowed human landing on the moon. On the other hand, the search for a cure for neurodegenerative conditions has been on the agenda from the second part of the past century. Vast amounts of intellectual, human, and financial resources have been spent—way more than what was spent on space missions but producing way fewer significant results.

    The use of the unbiased computational capability of AI now provides the tools to unravel the unique complexity of the "individual" disease status affecting each of our patients and allows the implementation of precision medicine-based approaches. The neurology of the future, a "2.0. neurology"—less obsessed with disease subclassifications and more devoted to generating tailored diagnoses and producing customized and effective therapies—will significantly benefit from a shift towards precision medicine, the use of unorthodox approaches, and the final departure from the post-Oslerian reductionism of the past.

    References:

    . Neurodegeneration. Science. 2020 Oct 2;370(6512):48-49. PubMed.

    . Revisiting protein aggregation as pathogenic in sporadic Parkinson and Alzheimer diseases. Neurology. 2019 Feb 12;92(7):329-337. PubMed.

    . Person-specific contribution of neuropathologies to cognitive loss in old age. Ann Neurol. 2018 Jan;83(1):74-83. Epub 2018 Jan 14 PubMed.

    . Attributable risk of Alzheimer's dementia attributed to age-related neuropathologies. Ann Neurol. 2019 Jan;85(1):114-124. Epub 2018 Dec 19 PubMed.

    . Targeted brain proteomics uncover multiple pathways to Alzheimer's dementia. Ann Neurol. 2018 Jul;84(1):78-88. Epub 2018 Jul 3 PubMed.

    . Prevalence and Clinical Phenotype of Quadruple Misfolded Proteins in Older Adults. JAMA Neurol. 2020 Jun 22; PubMed.

    . Systems biology and the future of medicine. Wiley Interdiscip. Rev. Syst. Biol. Med. 3. 2011

  3. It’s not over yet!

    We all wanted to believe that aducanumab would spell the beginning of the end of AD-related dementia. But the evidence that this promising anti-amyloid treatment might buck the unfortunate historical trend haunting anti-amyloid treatments is currently lacking: not only have methodological concerns been raised concerning the ENGAGE and EMERGE trials (Knopman et al., 2020), but effect sizes must also be taken into consideration.

    For comparison, the French government decided to stop reimbursing “insufficiently efficacious” cholinesterase inhibitors which, for tens of dollars, slow cognitive decline as measured by MMSE by 1.37 points over six months versus placebo (Birks, 2006). Thus, when Biogen/Eisai showed that high-dose aducanumab led to an 18 percent change in MMSE score, it was in fact a 0.6 point MMSE change over 18 months. Of course, this is “early stage AD” and could mean a larger difference over longer time periods. But this remains to be proven. And besides, who can afford tens of thousands of dollars a year for such a tiny long-term effect?

    So what now? We conceive three future possibilities, depending on how the community wishes to position itself with respect to current anti-amyloid approaches. First, unsuccessful trials do not imply that we can reject the role of β-amyloid in AD (Selkoe and Hardy, 2016). A first lesson might be to move toward early prevention of AD-related dementia with validated amyloid-removing treatments, acting within the right treatment time window (e.g., “killing the seeds” early on to prevent further accumulation of amyloid brain lesions as advocated by Schelle et al., 2019). Second, we might keep the same reductionist methodology, but change therapeutic target, for example, toward tau (see ongoing immunotherapy trials). Third, more radical changes could include more systemic approaches attempting to bridge the gap between promising lifestyle studies on prevention (Livingston et al., 2020) and knowledge of AD’s molecular biology (Cummings et al., 2020). A problem seems to be that as soon as we move away from amyloid-based reductionism, we move toward a kind of target anarchy, Herrup’s “choose them all” strategy (Herrup, 2015). 

    Our group argues that β-amyloid stands at the heart of pluralistic etiological paradigms since good evidence suggests that “When There’s no Amyloid, It’s not Alzheimer’s” (Sep 2015 news). We cannot, therefore, jump ship from amyloid-based approaches. Instead, we must integrate knowledge of other processes around this central process.

    Non-specific treatments will surely have their place within the therapeutic arsenal against AD, but designing effective therapies—and rigorous trials to test them—is proving almost impossible in this frail, treatment-resistant population, and “shotgun” therapy should not be seen as the alternative to targeted approaches. The effect of the COVID-19 pandemic on patient recruitment for clinical trials should also not be neglected.

    If no new drug is validated against AD in the coming years, more pharmaceutical companies will follow Pfizer’s 2018 high-profile exit from AD research. It is worth nothing that the Phase 3 aducanumab trials were halted prematurely for financial and not scientific reasons, before being resuscitated and presented (badly) to the scientific community for—once again—financial and not scientific reasons. They were finally halted—for scientific reasons this time—by the FDA advisory panel.

    Given that shareholders, not patient stakeholders, are at the center of the drug discovery process, if convincing pharmaceutical companies to keep studying AD becomes impossible, we may need to imagine new types of public-private ventures, as the conventional capitalist model of research against this devastating disease appears to be reaching tipping point.

    References:

    . Failure to demonstrate efficacy of aducanumab: An analysis of the EMERGE and ENGAGE trials as reported by Biogen, December 2019. Alzheimers Dement. 2021 Apr;17(4):696-701. Epub 2020 Nov 1 PubMed.

    . Cholinesterase inhibitors for Alzheimer's disease. Cochrane Database Syst Rev. 2006;(1):CD005593. PubMed.

    . The amyloid hypothesis of Alzheimer's disease at 25 years. EMBO Mol Med. 2016 Jun 1;8(6):595-608. PubMed.

    . Early Aβ reduction prevents progression of cerebral amyloid angiopathy. Ann Neurol. 2019 Oct;86(4):561-571. Epub 2019 Aug 19 PubMed.

    . Dementia prevention, intervention, and care: 2020 report of the Lancet Commission. Lancet. 2020 Aug 8;396(10248):413-446. Epub 2020 Jul 30 PubMed.

    . Alzheimer's disease drug development pipeline: 2020. Alzheimers Dement (N Y). 2020;6(1):e12050. Epub 2020 Jul 16 PubMed.

    . The case for rejecting the amyloid cascade hypothesis. Nat Neurosci. 2015 Jun;18(6):794-9. PubMed.

  4. We will see the results of further studies based on the amyloid hypothesis as studies continue to be pursued globally. We and others have suggested important caveats to interpreting clinical trial results. Broader thinking is now occurring across the Alzheimer's field. Ultimately a new unifying hypothesis of disease that accounts for all the major observations is needed. Please see the papers cited below.

    References:

    . Inconsistencies and controversies surrounding the amyloid hypothesis of Alzheimer's disease. Acta Neuropathol Commun. 2014 Sep 18;2:135. PubMed.

    . Questions concerning the role of amyloid-β in the definition, aetiology and diagnosis of Alzheimer's disease. Acta Neuropathol. 2018 Nov;136(5):663-689. Epub 2018 Oct 22 PubMed.

    . Microglia: A new frontier for synaptic plasticity, learning and memory, and neurodegenerative disease research. Neurobiol Learn Mem. 2013 Oct;105:40-53. PubMed.

  5. The negative vote on Biogen’s aducanumab filing by the FDA Advisory Committee constitutes a clear setback for future availability of this antibody to oligomeric Aβ for Alzheimer patients. The basis for this vote, and its implications, needs to be carefully considered. Listening to the meeting, I felt that the panel members were not given enough time to have all their questions addressed on the discrepancy in cognitive outcomes between the high-dose cohorts of trial 301 (ENGAGE) and 302 (EMERGE). In particular, no consensus was reached on how to calculate the significance statistic of 301 and 302 together, given that Biogen and the FDA proposed to focus on 302 (generally positive) and set aside 301 (negative), with the explanation that 301 subjects overall had less total brain exposure to aducanumab and therefore showed modestly less amyloid lowering and secondary CSF pTau lowering than the 302 subjects. This key concern about an explanation for the high-dose discrepancy did not undergo an adequate dialogue between the panel members and the sponsor and agency, and the reviewers were unpersuaded that this exposure difference explains the disparate trial outcomes.

    Because numerous physicians with experience in Alzheimer’s disease believe that there is some evidence of a positive signal in the 302 and 103 (Phase 1b) trials on both biomarkers (amyloid- and tau-lowering) and aspects of cognition (CDR-SB, MMSE, ADAS-cog 13, ADCS-ADL, and/or NPI-13), I believe that the imperfect discussion of 301 versus 302 last Friday should lead to renewed efforts to shed more light on the basis for the discrepant results.

    First, Biogen should move ahead quickly to subject all of the trial data to peer review to bring full transparency to all study outcomes and their statistical analyses. Data availability can be achieved relatively quickly by journal submission and simultaneous posting of the submitted papers at BioRxiv.org, as is now commonly done.

    Second, the company could choose to assay blindly all plasma samples from both trials for levels of the emerging tau analytes that track with progression of AD pathobiology, pT217 tau, and NT1 tau, looking for lower mean levels in trial subjects exposed to aducanumab versus placebo. This would obviously not qualify as a biomarker outcome, but a blinded analysis might provide further biological evidence that the agent benefited AD pathobiology in the trials and help illuminate whether high dose in 301 produced less such benefit than high dose in 302.

    Third, the FDA, in my view, should still carefully consider the apparently positive clinical and biological effects observed in 302. Though the effect size at 18 months may appear modest, a disease-modifying agent (it lowers both of the defining lesions of AD) would be expected to show greater benefit after longer dosing. Fourth, the FDA may consider any emerging data from the EMBARK open-label extension trial now underway to learn whether any benefits are reestablished and then maintained.

    The personal and societal tragedy of Alzheimer’s, and the great difficulty in finding agents that can move its brain pathology in association with some clinical benefit in just 18 months, suggest to me that our field should be cautious before setting aside an agent that has now been tested in thousands of patients and appears to confer some benefit in a subset, with manageable safety. Other, better agents will no doubt emerge in some years, but a number of well-informed AD physicians believe that aducanumab is capable of lowering Aβ oligomers in AD brain and therefore deserves further careful consideration by the FDA, perhaps including some form of “conditional approval.” We owe it to the myriad patients and families who have given so much to this effort to be very careful in deciding whether the totality of evidence on aducanumab predicts it could help some patients with this relentlessly progressive disease. Such therapeutic controversies are not uncommon in other chronic, fatal disorders.

  6. It was stimulating to read these assorted comments on the recent turndown of aducanumab by the FDA advisory committee. Another defeat, perhaps temporary, for the amyloid hypothesis of Alzheimer's. Most comments were devoted to salvaging it.

    But no other possible causes were mentioned as now deserving more urgent investigation. I put the words of the article and all comments into my Word program; the total word count was 3,000-3,500, depending on what one included. My positive control was the word "the," which appeared 271 times. The words "microbe," "bacteria," and "virus" were not detected even once. Perhaps we need to reconsider research allotments?

Make a Comment

To make a comment you must login or register.

References

Therapeutics Citations

  1. Aduhelm

News Citations

  1. Biogen/Eisai Halt Phase 3 Aducanumab Trials
  2. ‘Reports of My Death Are Greatly Exaggerated.’ Signed, Aducanumab
  3. Exposure, Exposure, Exposure? At CTAD, Aducanumab Scientists Make a Case
  4. Biogen Asks FDA To Approve Aducanumab
  5. Biogen Antibody Buoyed by Phase 1 Data and Hungry Investors
  6. Paper Alert: Aducanumab Phase 1b Study Published

Further Reading